Gold colloid changes its color when the internanoparticle distance changes. On the basis of analyte-induced aggregation or disaggregation behavior of gold nanoparticles (AuNPs), versatile colorimetric assays have been developed for measuring various kinds of analytes including proteins, DNA, small molecules, and ions. Traditional read-out signals, which are usually measured by a spectrometer or naked eyes, are based on the averaged extinction properties of a bulk solution containing billions of nanoparticles. Averaged extinction property of a large amount of nanoparticles diminished the contribution from rare events when the analyte concentration was low, thus resulting in limited detection sensitivity. Instead of measuring the averaged optical property from bulk colloid, in the present work, we proposed a digital counterpart of the colorimetric assay by imaging and counting individual AuNPs. This method quantified the analyte concentration with the number percentage of large-sized AuNPs aggregates, which were digitally counted with surface plasmon resonance microscopy (SPRM), a plasmonic imaging technique recently developed by us and other groups. SPRM was able to identify rare AuNPs aggregates despite their small population and greatly improved the detection sensitivity as demonstrated by two model systems based on analyte-induced aggregation and disaggregation, respectively. Furthermore, besides plasmonic AuNPs, SPRM is also suitable for imaging and counting nonplasmonic nanomaterials such as silica and metal oxide with poor extinction properties. It is thus anticipated that the present digitized assay holds a great potential for expanding the colorimetric assay to broad categories of nonplasmonic nanoparticles.